Understanding Mixed-Mode Retention Mechanisms in Liquid

Apr 11, 2014 - Wyndham , K. D.; O'Gara , J. E.; Walter , T. H.; Glose , K. H.; Lawrence , N. L.; Alden , B. A.; Izzo , G. S.; Hudalla , C. J.; Iraneta...
0 downloads 0 Views 533KB Size
Download PDF
Article pubs.acs.org/ac

Understanding Mixed-Mode Retention Mechanisms in Liquid Chromatography with Hydrophobic Stationary Phases Alberto Cavazzini,*,† Nicola Marchetti,† Roberta Guzzinati,† Luisa Pasti,† Alessia Ciogli,‡ Francesco Gasparrini,‡ and Aldo Laganච†

Department of Chemistry and Pharmaceutical Sciences, University of Ferrara, via L. Borsari, 46, 44121 Ferrara, Italy Dipartimento di Chimica e Tecnologia del Farmaco, University “La Sapienza”, 00185 Rome, Italy ¶ Department of Chemistry, University of Rome, “La Sapienza”, Piazzale Aldo Moro 5, 00185 Rome, Italy ‡

ABSTRACT: The chromatographic retention mechanisms of two hydrophobic bonded phases, octadecyl ethyl-bridged organic/inorganic (BEH-C18) and straight-chain perfluorohexylpropyl silica (C6F13), have been investigated by using a homologous series of alkyl-benzenes and perfluoroalkyl acids as test compounds in a variety of acetonitrile/water mobile phases and at different temperatures. On both columns, polar compounds exhibited a characteristic Ushape retention behavior in function of acetonitrile amount in the eluent, whereas retention of neutral molecules decreased continuously, following an increase of organic modifier, over the entire mobile phase range. The dependence of perfluoromethylene selectivity upon eluent composition explains the typical reversed-phase behavior (decreasing in retention following an increase of acetonitrile in mobile phase) initially exhibited by perfluoroalkyl acids, but alone it cannot justify their increasing of retention at organic-rich mobile phases (approximately >90% v/v for acetonitrile with the C6F13 column and acetonitrile >80% v/v for the BEH-C18 one). It actually predicts an opposite trend, indicating thus the presence of mixedmode retention mechanisms. Indeed it was found that, at organic-rich mobile phases, the transfer from the mobile to the stationary phase of the polar moiety of molecules drives retention. This finding has been correlated to the excess adsorption isotherm of acetonitrile/water binary mixtures and thus to the composition of the stationary phase. At organic-rich mobile phases, in fact, stationary phases are characterized by a positive excess of adsorbed water that creates an “environment” suitable to the transfer herein of polar groups.

R

discriminate the relative hydrophobicity of different bonded phases.3,4 There are, however, several findings showing that this simple model does not take into account the intrinsic complexity of RP chromatography. For instance, it has been reported by many authors that polar compounds, under certain conditions,5−21 exhibit a characteristic U-shaped retention profile: by increasing the organic modifier (very often acetonitrile) in the mobile phase, their retention decreases only up to a certain point where, at organic-rich eluents, it begins to increase. The point of minimum retention is however unpredictable a priori and changes with the characteristics of adsorbent−eluent system. These findings suggest, and have been interpreted in terms of, more complex retention mechanism where solute retention is postulated to be a result of both hydrophobic and silanophilic interactions. The latter are due to unbonded/unend-capped silica surface acidic silanols that are present on large amounts on silica-based RP packings. It is indeed known that, due to steric hindrance, only roughly half out of the about 8 μmol/m2, which

eversed-phase (RP) chromatography is by far the most commonly used mode in high-performance liquid chromatography (HPLC). It uses hydrophobic packings, very often silica-based bonded phases (where the ligand is a longchain hydrocarbon), and polar eluents usually made of mixtures of water or buffer with polar solvents such as acetonitrile, methanol, or tetrahydrofuran. The rule of thumb to explain retention under RP conditions is that the hydrophobic adsorbent interacts with the hydrophobic portion of analytes. The larger this portion, the stronger the retention; vice versa, the greater the mobile phase solubility of analytes, the smaller is the retention (from this it follows the empirical rule that in RP cromatography, retention decreases by increasing the amount of organic modifier in the mobile phase).1,2 The linear dependence of the logarithm of the retention factor upon the number of carbons in the backbone chain of homologous series (e.g., methylene groups for alkyl homologues) at fixed mobile phase composition is a good example of the correlation between retention and hydrophobic character of molecule.3 The slope of these plots allows for the calculation of the chromatographic selectivity (in the case of alkyl homologues, it is said methylene selectivity), which indeed has been used to © 2014 American Chemical Society

Received: January 20, 2014 Accepted: April 11, 2014 Published: April 11, 2014 4919

dx.doi.org/10.1021/ac500248b | Anal. Chem. 2014, 86, 4919−4926

Analytical Chemistry

Article

C18) and a straight-chain perfluorohexylpropyl silica (C6F13). To this end, homologous series of neutral (alkyl-benzenes) and polar (perfluoroalkyl acids) compounds have been employed. The retention behavior of these molecules was investigated as a function of eluent composition (water/acetonitrile binary mixtures to which 0.1% v/v formic acid was added) and temperature. Through this information, the contributions to retention coming from the different parts (polar and hydrophobic) of molecules could be singled out.20 Straight-chain perfluoralkyl stationary phases are considered to be more hydrophobic than their alkyl (C8, C18) counterparts.44,45 BEHC18 adsorbents are claimed to be, on the one hand, more stable and durable at high temperature and harsh pH conditions than traditional octadecyl-functionalized silica gels while possessing, on the other hand, very similar selectivity.46−48 For these characteristics, in this work, we have preferred to use a BEH-C18 column rather than a traditional C18 silica-based phase. On both phases, the Gibbs free energy change for the transfer from the mobile to the stationary phase of the carboxylic moiety was found to change sign from positive to negative (that is, thermodynamically speaking, from unfavorable to favorable to retention) in correspondence of the zones of the excess isotherm (measured under the same experimental conditions) where also the excess of acetonitrile changes from positive to negative. The postulated hydrophobic/silanophilic dual retention mechanism thus finds a straightforward interpretation in terms of stationary phase composition/properties of solutes. This work has also allowed for the characterization of stationary phases in terms of methylene/perfluoromethylene selectivity and as capacity of adsorbing acetonitrile from binary aqueous−organic mixtures. This last information has a significant practical relevance in an apparently different field of application, such as the fluorotelomer industry. The characterization of the stability of fluorotelomer polymers (a very actual and challenging analytical problem49) requires in fact, as basic information, knowledge of the “affinity” of organic solvents toward fluorotelomers and perfluorinated monomer compounds.

represents the accepted value for the number of silanols present on the surface of a fully hydroxylated bare silica,22 can be reacted during functionalization/end-capping reactions.2,23,24 The liquid/solid phase boundary region (also called interfacial layer or surface phase) is a complex and dynamic zone extending for a few molecular diameters from the surface of the adsorbent. In RP liquid chromatography with binary aqueous−organic mobile phases, due to surface attractive forces, this region is enriched in one of the components of the mixture (the so-called preferentially adsorbed or strong-modifier) and impoverished in the other (because there are not vacancies in the surface phase nor in the bulk solution, the preferential adsorption of one component of the mixture at the boundary region provokes the displacement of the other). Fundamental studies on adsorption have shown that the local concentrations of adsorbable components decrease progressively with increased distance from the adsorbent surface (some authors indeed speak of a “continuity of stationary phases, each at a given distance from the surface and with its own composition”25). This makes the distinction between mobile and stationary phases in RP chromatography difficult and not clear-cut.20,25−30 Things are further complicated by the fact that the equilibrium composition of solvents in the interfacial region depends, in a quite complex manner, on the bulk mobile phase composition.25,28−30 Adsorbate density and composition profiles within the interfacial layer cannot be measured by today’s technology. However, the excess sorption isotherm for the solvents−chromatographic packing system of interest can be straightforwardly experimentally determined through several direct dynamic techniques such as, for instance, the tracer-pulse or the perturbation on the plateau chromatography.30−35 More recently, molecular simulation studies have also been demonstrated to be very promising tools to model the interface space and its composition as a function of surrounding medium.36−40 To deal with this complexity, the concept of the Gibbs dividing surface (GDS) has been introduced as a hypothetical geometrical surface with a determined position that represents the boundary between the two phases.26,29 The position of the GDS can be defined on the basis of a given convention. Two conventions are commonly used in liquid chromatography for establishing the position of GDS: either the GDS is assumed to coincide with the physical surface of the solid adsorbent (so-called “nothing is adsorbed in terms of volume”, or v-NA, convention27) or the GDS is established at the interface between an adsorbed film (that constitutes the stationary phase volume) and the bulk fluid (this convention is said “component J not adsorbed”, or J-NA, convention27). With v-NA convention, the void volume of the column (also referred to as “thermodynamic dead volume”)28−30 is given by the total volume of the eluent in the column, which is commonly estimated by weighing the column before and after flushing it with a liquid of known density (pycnometry). If the JNA convention is adopted, the pertinent void volume is the sodefined “kinetic void volume”.25,28−30 This last is usually determined by injecting a compound that is believed not to be adsorbed (i.e., not entering the adsorbed layer) nor excluded (such as uracil in RP chromatography).29,41,42 Another approach to estimate the kinetic void volume is based on the analysis of the excess isotherm course and, in particular, of the region of the isotherm corresponding to the saturation of the stationary phase, as first described by Schay and Nagy.43 In this paper, we present a study aimed at investigating the retention mechanisms on two hydrophobic stationary phases, namely, a C18-ethyl-bridged hybrid organic/inorganic (BEH-

■

THEORY For the sake of concision, only a summary of fundamental equations employed in this work is presented. Some are the traditional equations used in linear chromatography for the estimation, for instance, of retention factor, eq 1, selectivity, eq 2, or the van’t Hoff equation, eq 4. As for the determination of thermodynamic quantities by tracer pulse chromatography, eqs 8−10, readers are referred to specific literature for more detailed information.20,30−34,50 Traditional Relationships in Linear Chromatography. The retention factor is defined as k=

VR − VM VM

(1)

where VR is the solute retention volume and VM the kinetic void volume. The selectivity or separation factor, α, is defined as the relative retention calculated for two adjacent peaks:51

α=

k2 k1

(2)

with k2 and k1 defined as the retention factors for the more and the less retained compound, respectively. With homologous series, α is best calculated by the slope of ln k versus carbon 4920

dx.doi.org/10.1021/ac500248b | Anal. Chem. 2014, 86, 4919−4926

Analytical Chemistry

Article

number in the backbone chain plot.3 Depending on the homologues, different selectivities can be defined, such as methylene selectivity, αCH2, for alkyl homologues or perfluoromethylene selectivity, αCF2, for perfluoroalkyl homologues (e.g., perfluoroalkyl acids).20,52−54 The natural logarithm of methylene or perfluoromethylene selectivity multiplied by the factor −RT (R being the universal gas constant and T the temperature in Kelvin) gives the change of Gibbs free energy for the transfer from the mobile to the stationary phase respectively of a methylene or a perfluoromethylene group: ◦ −RT ln αCX 2 = ΔGCX X = H, F 2

where VS (i.e., the negative of the slope of the linear regression) and VSACN (intercept) are the volume of the stationary phase and the volume of acetonitrile in the stationary phase, respectively.20,30,42 Herefrom, the kinetic void volume, VM, can be readily calculated, being

VM = V0 − VS so that also F, eq 5, can be estimated.

■

EXPERIMENTAL SECTION AND METHODS Columns and Chemicals. Two columns were employed: (1) Fluophase-RP C6F13 (perfluorohexylpropylsiloxane-bonded silica), 50 × 2.1 mm, 5 μm particle size, 100 Å pore size (from Thermo Scientific); (2) Acquity UPLC BEH-C18, 100 × 2.1 mm, 1.7 μm particle size, 130 Å pore size (from Waters Corp.). Perfluoroalkyl acids (perfluoropentanoic, perfluorohexanoic, perfluoroheptanoic, and perfluorooctanoic) and alkyl benzenes (benzene, toluene, ethylbenzene, propylbenzene, butylbenzene, pentylbenzene) were purchased by Sigma-Aldrich. Ultrahigh quality milli-Q water was obtained by a milli-Q water purification system (Millipore). Acetonitrile was LC/MS grade. Deuterated compounds (water and acetonitrile) were from Cambridge Isotope Laboratories, Inc. Formic acid was from Sigma-Aldrich. Equipment and Chromatographic Conditions. A LC/ MS instrument composed by a micro-HPLC (Finnigan Surveyor Plus) and a LTQ-XL linear ion trap MS detector (Thermo Scientific) was employed for tracer pulse chromatography (with an APCI ion source) and detection of perfluoroalkyl acids (ESI ion source). An Agilent 1290 (standard configuration) with UVDAD detection was employed for the determination of alkylbenzenes (wavelength, 260 nm). For tracer-pulse experiments, 5 μL injections of deuterated acetonitrile and deuterated water were performed independently. The column was equilibrated at each aqueous/acetonitrile mobile phase composition with at least 10 column volumes before injection. Acetonitrile concentration (v/v) was varied with increments of 10% in the range of 0−80%. Between 80 and 100%, the following concentrations were prepared: 85, 90, 93, 95, 97, and 100%.20 Measurements were done in triplicate. Retention times of perturbations were determined through peak moments. For van’t Hoff plots, retention data were collected at six different temperature (10, 20, 30, 40, 60 °C). The temperature was controlled (±0.1 °C) by a digital contact thermometer (IKA Lab Equipment). The flow rate was 0.3 mL/min for retention studies of homologues series (alkyl and perfluoroalkyl) and 0.1 mL/min during tracer pulse chromatography experiments. Peak’s first statistical moment was employed for the determination of retention time. All measurements were performed as triplicate determinations.

(3)

where X = H for alkyl homologues and X = F for perfluoroalkyl ones. Finally, the van’t Hoff relationship correlates k to the enthalpy and entropy change of the transfer of solute from the mobile to the stationary phase (ΔH° and ΔS°, respectively):31 ln k = −

ΔH ° ΔS° + + ln F RT R

(4)

F is the phase ratio, given by F = VS/VM

(5)

with VS as the volume of the stationary phase. From eq 4, granted the knowledge of F and the linearity of van’t Hoff plot,55 the total free energy ΔG0 for the transfer of a molecule from the mobile to the stationary phase can be calculated with the following equation: ΔG° = ΔH ° − T ΔS °

(6)

Frequently, molecules possess different functional groups with different polarity and interactive characteristics. For relatively simple molecules, following Martin56 it can be assumed that each group l of the molecule is associated with its own unique change ΔG0l in free Gibbs energy, independent of the presence of other groups, so that ΔG0 can be written as the linear combination of them:20,37,57,58 ΔG° =

∑ ΔGl0 l

(7)

Tracer Pulse Chromatography for the Determination of Excess Isotherm, VM and VS. The theory of tracer pulse chromatography applied to a (i, j) binary system28,30−34,42,59−61 shows that the excess volume of the isotopically labeled i 30 compound, Vexc i can be calculated by * i − V R, * j)θiMθjM V iexc = (V R,

■

(8)

RESULTS AND DISCUSSION Figure 1 shows how the excess volume of acetonitrile changes as a function of θM ACN· In the upper part of the figure (Figure 1a), data for the C18−BEH column are presented; the lower part of the same figure (Figure 1b) shows the variation of Vexc ACN in the C6F13 column. As already mentioned, the octadecyl ethyl-bridged organic−inorganic column was preferred to a traditional octadecyl silica-based column only in virtue of its greater stability at elevated temperature and harsh pH conditions.46 Qualitatively, the behavior of the two isotherms is very similar. They are characterized by a so-called S-shape due to a change of M sign of Vexc ACN at a given θACN (the point where excess is zero is referred to as adsorption azeotrope).43,62 Initially, the excess

where VR,i * and VR,j * are the elution volumes for each labeled M component i and j and θM i and θj their volume fractions in the bulk mobile phase. With the same set of experimental data, it is also possible to estimate the thermodynamic void volume, V0, by * iθiM + V R, * jθjM V0 = V R,

(9)

Following Schay and Nagy,43 the capacity and thickness of the surface phase can be estimated by linear fitting of the excess exc isotherm region where the excess of acetonitrile (VACN ) M decreases by increasing its volume fraction (θACN), being exc S M V ACN = V ACN − VSθACN

(11)

(10) 4921

dx.doi.org/10.1021/ac500248b | Anal. Chem. 2014, 86, 4919−4926

Analytical Chemistry

Article M volume of acetonitrile increases by increasing θACN , as a consequence of the adsorption by hydrophobic phases. Then, after passing through a maximum, the excess decreases quasilinearly with θM ACN. It is where saturation of the adsorbent has been reached43,63,64 and the composition and volume of the stationary phase (i.e., VSACN and VS, see eq 10) remain constant and independent of θM ACN. This region (saturation zone) is particularly important for studying retention mechanisms. In fact, because it can be assumed that the very low amounts injected in analytical chromatography do not alter the stationary phase composition,16 a change of retention following a change of θM ACN can be considered, in this region, to be a function only of the mobile phase variation. On the other hand, the same reasoning cannot be applied to the initial part of the isotherm (i.e., for θM ACN < 0.45), as a change in θM ACN, therein also provokes a variation of the volume and composition of the stationary phase. The application of eq 10 to the region of linear decrease of isotherms (see Figure 1) allows for the estimation of VSACN and VS. Through linear fitting, the following values were obtained: VSACN = 0.029 mL and VS = 0.036 mL for the BEH-C18 column, and VSACN = 0.023 mL and VS = 0.025 mL for the C6F13 one. At saturation, therefore, roughly 80% in volume of the stationary phase of the BEH-C18 column is made of acetonitrile. This percentage grows up to more than 90% with the perfluorinated adsorbent. As thermodynamic void volumes can be calculated through eq 9 (V0 = 0.22 was obtained for the BEH-C18, and V0 = 0.13 mL for the C6F13), estimation of kinetic void volumes (eq 11) and phase ratios (eq 5) is straightforward in this region (VM = 0.19 and F = 0.20 were calculated for the BEH-C18 column, whereas VM = 0.10 mL was measured on the C6F13 column with F = 0.24). As depicted in Figure 1, the amplitude of zones where acetonitrile excess is negative is different on the two columns (so is different the position of isotherm minimum), reflecting different abundance/accessibility of surface unreacted silanols.65 This region is wider on the octadecyl-silica column (approximately 0.8 < θM ACN < 1) than on the perfluorinated one (0.95
0.95 (C6F13 column) or θM >0.8 (BEH-C column), ln k ACN 18 increases with θM ACN thus, with a typical hydrophilic chromatography- (HILIC)-like behavior.66 In the HILIC region, the difference in retention among corresponding perfluoroalkyl acids on the two columns, which was practically constant in the RP region, decreases markedly. By taking PFOA as an example, this difference reduces by approximately two logarithmic units by going from θM ACN = 0.5 to 0.95. Simultaneously, in this region, the selectivity of both columns toward the separation of acids reduces (more markedly on the BEH-C18 column, where at θM ACN = 1 is practically lost). In a subsequent section in this mansucript, a thermodynamically based explanation to the observed selectivity loss will be proposed. From slopes of ln k versus the number of carbons in the 3,20 backbone chain plots, at constant θM ACN, αCF2 can be estimated. In Table 1, second and third columns, perfluoromethylene

refer to the C6F13 stationary phase, and black circles refer to the BEH-C18). There are some interesting things that can be noticed by looking at this figure. First of all, it can be observed that on both columns ΔG0CF2 is negative over the entire mobile phase range. Therefore, the transfer of a CF2 unit is always thermodynamically favorable to retention.37,57,67 By increasing the amount of organic in the mobile phase, however, this transfer becomes progressively less favorable (the absolute value of ΔG0CF2 becoming smaller). For both columns, the plot ΔG0CF2 versus θM ACN is characterized by a wide zone where the slope of ΔG0CF2 is nearly constant. These zones correspond to the saturation regions of the respective excess isotherms, where the composition and volume of stationary phases are also constant (see Figure 1), and the retention of perfluoroalkyl acids exhibit the typical RP behavior (see Figure 2). The constant increase of ΔG0CF2 with θM ACN in these zones can be explained in terms of an improved solubility of perfluoroalkyl acids in eluents increasingly richer in organic character; on the other hand, the composition of the stationary phase, and so the solubility of acids herein, remains constant. By consider Figure 3 with respect to mobile phase compositions either poorer or richer in acetonitrile than those of the RP region, one observes that, in both cases, the slope of ΔG0CF2 increases. The explanation of this increase is however different in the two cases. In the former (θM ACN smaller than roughly 50% v/v), the stationary phase has not been yet saturated by acetonitrile, and its volume and composition changes with 0 M θM ACN. The steeper dependence of ΔGCF2 on θACN in this region can be explained by considering the concomitant effect of two phenomena:63 the reduction of solubility of perfluorinated molecules in water-rich mobile phases adds up, indeed, to the presence of hydrophobic interactions between analytes and alkyl or perfluoroalkyl chains tethered to the surface, interactions that are more effective when the stationary phase has not been saturated by acetonitrile. In confirmation of this hypothesis, one may observe that the distance between empty and black circles (Figure 3), at each mobile phase composition, visually represents the different perfluoroselectivity of stationary phases increases in this region, as fluorine−fluorine interactions are known to be stronger and more selective than those of fluorine−hydrogen.20,68 Finally, the increase of slope in the acetonitrile-rich region (HILIC zone), which determines a less favorable transfer of the CF2 unit with increasing θM ACN, originates by the fact that herein stationary phases of both adsorbents are characterized by an excess of water (see Figure 1), which reduces the solubility of perfluoroalkyl chains. The results obtained about the dependence of ΔG0CF2 on θM ACN do not allow an explanation of the U-shape retention observed for perfluoro-alkyl acids (Figure 2). Indeed, based on ΔG0CF2 values, one would expect retention of perfluoroalkyl acids to decrease constantly (even though not linearly) by increasing θM ACN, which is evidently an erroneous conclusion. The reason for this apparent contradiction is that, so far, only the contribution to retention coming from the transfer of the perfluorinated portion of molecules was considered. Perfluorinated acids, however, also possess an hydrophilic moiety that does contribute to their chromatographic behavior.

Table 1. Dependence of Perfluoromethylene (Second and Third Columns) and Methylene (Fourth and Fifth Columns) Selectivities on the Mobile Phase Composition for the Two Stationary Phases Considered in This Worka αCF2

αCH2

θM ACN

C6F13

BEH-C18

C6F13

BEH-C18

0.4 0.45 0.5 0.6 0.7 0.8 0.9 0.95 0.97

NA 3.15 2.47 2.25 2.05 1.97 1.78 1.72 1.46

2.02 1.98 1.76 1.63 1.53 1.44 1.36 1.25 1.22

1.53 1.43 1.36 1.27 1.21 1.19 1.12 NA NA

1.86 1.75 1.68 1.56 1.47 1.39 1.31 NA NA

a

Symbol NA (not applicable) indicates cases where experimental data were not measured (for perfluoroalkyl acids, as retention was excessively large; for alkyl benzenes, as data were unnecessary in the series).

selectivities calculated for the two columns at different mobile phase compositions are reported. These values are the basis for the calculation of ΔG0CF2 (eq 3), which is the Gibbs free energy change for the transfer from the mobile to the stationary phase of a perfluoromethylene unit. The dependence of ΔG0CF2 on θM ACN has been graphically shown in Figure 3 (left y-axis) for both stationary phases (empty circles 4923

dx.doi.org/10.1021/ac500248b | Anal. Chem. 2014, 86, 4919−4926

Analytical Chemistry

Article

Figure 3. Free energy changes for the transfer of a CF2 and a COOH unit from the mobile to the stationary phase as a function of the mobile phase composition, at room temperature (298 K). Left y-axis: ΔG0CF2 for the C6F13 (○) and the BEH-C18 (●) column. Right y-axis: ΔG0COOH for the C6F13 (□) and the BEH-C18 (■) column.

Figure 4. Retention behavior of alkyl-benzenes on the C6F13 column as a function of the mobile phase composition. □: benzene; ■: toluene; ○: ethylbenzene; ·: propyl-benzene; Δ: butyl-benzene; ▲: pentyl-benzene.

essentially consists of the following: (i) obtaining ΔH0 and ΔS0 by the dependence of ln k on the inverse of T (eq 4) at each mobile phase composition (van’t Hoff’s plots were found to be linear, data not presented; in these calculations, as a first approximation, for each column, F was assumed to be constant even for mobile phase compositions not belonging to the saturation zone); (ii) calculating ΔG0 through eq 6; and finally, (iii), for each perfluoroalkyl acid estimating ΔG0COOH as (see eq 12):

Accordingly, the total free energy change for the transfer of the entire molecule from the mobile to the stationary phase can be written in terms of Martin’s model (eq 7), as 0 0 ΔG° = ΔGCOOH + nCF2 × ΔGCF 2

(12)

where ΔG0COOH is the free energy change associated with the transfer of the polar moiety of the molecule and nCF2 the carbon number in the backbone chain of perfluoroalkyl acids.20 The contribution to retention coming from the polar part can be singled out, following the procedure described in ref 20. This 4924

dx.doi.org/10.1021/ac500248b | Anal. Chem. 2014, 86, 4919−4926

Analytical Chemistry 0 0 ΔGCOOH = ΔG° − nCF2 × ΔGCF 2

Article

conditions is due to the effect of surface unreacted silanols. However, apolar molecules are only slightly influenced by these conditions (even though their transfer to the stationary phase is less favorable if an excess of water is present therein). When the content of water in the mobile phase exceeds approximately 10% v/v (for the perfluorinated stationary phase) or 20% v/v (for the octadecyl hybrid organic/inorganic phase), the stationary phase is instead characterized by an excess of acetonitrile. Under these conditions, retention is expected to decrease with increasing acetonitrile amount in mobile phase (thus, with a typical RP behavior). However, this dependence is not constant, and only if the stationary phase is saturated by acetonitrile (i.e., under condition of constant stationary phase composition) will we then expect the linearity (or quasi-linearity) of logarithm of retention factor versus composition of mobile phase plots. This study has also shown the elevated capacity of perfluorinated stationary phases to adsorb acetonitrile. This base information has a practical utility in fields such as the determination of perfluorinated emerging contaminants from environmental matrices and for studies on the stability of industrial fluorotelomer products.

(13)

The dependence of ΔG0COOH on θM ACN has been graphically represented in Figure 3 (right y-axis). In this plot, empty squares refer to the C6F13 column, whereas black ones are for the BEHC18 column. Incidentally, ΔG0COOH values reported in Figure 3 are the average of the four ΔG0COOHs calculated at constant θM ACN for each perfluoroalkyl acid.20 Error bars have not been shown, because they were practically invisible on the ΔG0COOH energy range (±4000 J mol−1) of Figure 3. Expressed as ±1 standard deviation, errors were roughly in the order of ±60−80 J mol−1. The most interesting result of this study is the abrupt change in the sign of ΔG0COOH, from positive to negative (therefore, thermodynamically speaking, from unfavorable to favorable to retention37,57,67) in correspondence, on both columns, to the region where the excess of acetonitrile becomes negative (HILIC-zone). The dashed lines connecting the experimental points have been used to emphasize this change. This change is larger on the BEH-C18 column than on the C6F13 one, and it seems thus to be consistent with the larger excess of water found on the octadecyl phase (Figure 1). It could also explain why the selectivity of the BEH-C18 column toward the separation of perfluoroalkyl acids in the HILIC region is practically lost (see Figure 2). On the other hand, ΔG0COOH is substantially constant and independent of the mobile phase composition in the RP region. Together with the linear trend exhibited by ΔG0CF2 (Figure 3), this finding further explains the linearity of ln k versus θM ACN plots in the RP zone (Figure 2). In the final part of this work, the behavior of columns toward the separation of a series of alkyl homologues was considered. Not possessing any polar moiety, we expected the U-shaped retention trend to be lost with these compounds. The dependence of ln k on θM ACN for alkyl benzenes on the C6F13 column is shown in Figure 4. Retention decreases constantly by increasing θM ACN over the all mobile phase composition range. Looking deeper into the figure, the effect of the different zones of the adsorption isotherm (Figure 1) on retention of alkyl homologues, however, can be still recognized, even if in a clearly less evident manner than for perfluoroalkyl acids. Slopes of plots, in fact, become slightly steeper when θM ACN diminishes below 0.5 and they flatten out in the organic-rich zone (where, in addition, separation selectivity is practically lost). A quite analogous behavior was observed for the elution of alkyl homologues on the BEH-C18 column (data not presented). In this case, however, the retention of corresponding alkyl benzenes was systematically larger than on the C6F13, in confirmation of the general principle “like attracts like” (Table 1, fourth and fifth columns, lists the methylene selectivity calculated on the two columns).

■

AUTHOR INFORMATION

Corresponding Author

*E-mail: [email protected]. Fax: +39 0532 240709. Tel.: +39 0532 455331. Notes

The authors declare no competing financial interest.

■

ACKNOWLEDGMENTS The authors thank the Italian University and Scientific Research Ministry PRIN 2012ATMNJ_003, the University of Ferrara (Progetto Internazionalizzazione 2011, FAR 2012) and the Regional Operational Programme of the European Regional Development Fund, Industrial Research and Technological Transfer (POR FESR 2007-2013 Priority 1).

■ ■

DEDICATION Dedicated to Professor Francesco Dondi on the occasion of his retirement. REFERENCES

(1) Snyder, L. R.; Kirkland, J. J.; Dolan, J. W. Introduction to Modern Liquid Chromatography, 3rd ed.; Wiley & Sons: New York, 2010. (2) Neue, U. D. HPLC Columns: Theory, Technology, and Practice; Wiley-VCH: New York, 1997. (3) Bhagwat, V.; Bereznitski, T.; Buszewski, B.; Jaroniec, M. J. Liq. Chromatogr. Relat. Technol. 1998, 21, 923−939. (4) Sadek, P. C.; Carr, P. W.; Russo, M. J. Anal. Chem. 1987, 59, 1032− 1039. (5) Sadek, P. C.; Carr, P. W. J. Chromatogr. Sci. 1983, 21, 314−320. (6) Nawrocki, J.; Buszewski, B. J. Chromatogr. 1988, 449, 1−24. (7) Blanquet, R. S.; Bui, K. H.; Armstrong, D. W. J. Liq. Chromatogr. 1986, 9, 1933−1949. (8) Armstrong, D. W.; Boehm, R. E. J. Chromatogr. Sci. 1984, 22, 378− 385. (9) Simpson, R. J.; Moritz, R. L. J. Chromatogr. 1989, 474, 418−423. (10) Purcell, A. W.; Zhao, G. L.; Aguilar, M. J.; Hearn, M. T. W. J. Chromatogr. A 1999, 852, 43−57. (11) Purcell, A. W.; Aguilar, M. J.; Hearn, M. T. W. Anal. Chem. 1993, 65, 3038−3047. (12) Lee, D. W.; Cho, B. Y. J. Liq. Chromatogr. 1994, 17, 2541−2558. (13) Krause, E.; Beyermann, M.; Dathe, M.; Rothemund, S.; Bienert, M. Anal. Chem. 1995, 252−258.

■

CONCLUSIONS The detailed characterization of the composition of the interfacial region between the bulk solution and the adsorption surface and of how it changes as a function of the mobile phase composition has been the key to explain and, to a certain degree, even to predict the dependence of analyte retention upon changes of experimental conditions. In particular, it was demonstrated that the increase of retention of polar compounds at organic-rich mobile phases (U-shape behavior) is driven by the thermodynamically favorable transfer of the polar part of molecules from the mobile to a water-rich stationary phase. The excess of water present on the stationary phase under these 4925

dx.doi.org/10.1021/ac500248b | Anal. Chem. 2014, 86, 4919−4926

Analytical Chemistry

Article

(14) Lin, F.; Chen, W.; Hearn, M. T. W. J. Mol. Recognit. 2002, 15, 55− 93. (15) Nahum, A.; Horváth, C. J. Chromatogr. 1981, 203, 53−63. (16) Jaroniec, M. J. Chromatogr. A 1993, 656, 37−50. (17) Bell, D. S.; Jones, A. D. J. Chromatogr. A 2005, 1073, 99−109. (18) Bij, K. E.; Horváth, C.; Melander, W. R.; Nahum, A. J. Chromatogr. A 1981, 203, 65−84. (19) Tramposch, W. G.; Weber, S. G. Anal. Chem. 1984, 56, 2567− 2571. (20) Marchetti, N.; Caciolli, L.; Laganà, A.; Gasparrini, F.; Pasti, L.; Dondi, F.; Cavazzini, A. Anal. Chem. 2012, 84, 7138−7145. (21) Zhang, W. J. Fluorine Chem. 2008, 129, 910−919. (22) Zhuravlev, L. T. Langmuir 1987, 3, 316−318. (23) Cavazzini, A.; Pasti, L.; Greco, R.; Costa, V.; Solera, D.; Dondi, F.; Marchetti, N.; Laganà, A.; Gasparrini, F. J. Chromatogr. A 2013, 1286, 47−54. (24) Cavazzini, A.; Marchetti, N.; Pasti, L.; Greco, R.; Dondi, F.; Laganà, A.; Ciogli, A.; Gasparrini, F. Anal. Chem. 2013, 85, 19−22. (25) Poppe, H. J. Chromatogr. A 1993, 656, 19−26. (26) Gibbs, J. W. The Collected Works of J. W. Gibbs; Longmans, Green and Co.: New York, London, Toronto; Vol. I, Chapter III, 1928; p 219. (27) Riedo, F.; Kováts, E. S. J. Chromatogr. 1982, 239, 1−28. (28) Knox, J. H.; Kaliszan, R. J. Chromatogr. 1985, 349, 211−234. (29) Yun, K. S.; Zhu, C.; Parcher, J. F. Anal. Chem. 1995, 67, 613−619. (30) Mallette, J.; Wang, M.; Parcher, J. F. Anal. Chem. 2010, 82, 3329− 3336. (31) Fornstedt, T. J. Chromatogr. A 2010, 1217, 792−812. (32) Arnell, R.; Fórsen, P.; Fornstedt, T. J. Chromatogr. A 2005, 1099, 167−174. (33) Lindholm, J.; Forssen, P.; Fornstedt, T. Anal. Chem. 2004, 4856− 4865. (34) Lindholm, J.; Forssen, P.; Fornstedt, T. Anal. Chem. 2004, 5472− 5478. (35) Guiochon, G.; Felinger, A.; Shirazi, D. G.; Katti, A. M. Fundamentals of Preparative and Nonlinear Chromatography, 2nd ed.; Elsevier B.V.: The Netherlands, 2006. (36) Rafferty, J. F.; Zhang, L.; Siepmann, J. I.; Schure, M. R. Anal. Chem. 2007, 79, 6551−6558. (37) Rafferty, J. L.; Siepmann, J. I.; Schure, M. R. Adv. Chromatogr. 2010, 48, 1−55. (38) Lindsey, R. K.; Rafferty, J. L.; Eggimann, B. L.; Siepmann, J. I.; Schure, M. R. J. Chromatogr. A 2013, 1287, 60−82. (39) Melnikov, S. M.; Höltzel, A.; Seidel-Morgenstern, A.; Tallarek, U. Anal. Chem. 2011, 83, 2569−2575. (40) Melnikov, S. M.; Höltzel, A.; Seidel-Morgenstern, A.; Tallarek, U. Angew. Chem., Int. Ed. 2012, 51, 6251−6254. (41) Rimmer, C. A.; Simmons, C. R.; Dorsey, J. D. J. Chromatogr. A 2002, 965, 219−232. (42) Gritti, F.; Kazakevich, Y.; Guiochon, G. J. Chromatogr. A 2007, 1161, 157−169. (43) Schay, G.; Nagy, L. J. Chim. Phys. 1961, 140, 149−158. (44) Schoenmakers, P. J.; Billiet, H. A. H.; de Galan, L. Chromatographia 1982, 15, 205−214. (45) Billiet, H. A. H.; Schoenmakers, P. J.; de Galan, L. J. Chromatogr. 1981, 218, 443−454. (46) Wyndham, K. D.; O’Gara, J. E.; Walter, T. H.; Glose, K. H.; Lawrence, N. L.; Alden, B. A.; Izzo, G. S.; Hudalla, C. J.; Iraneta, P. C. Anal. Chem. 2003, 6781−6788. (47) Samuelsson, J.; Franz, A.; Stanley, B. J.; Fornstedt, T. J. Chromatogr. A 2007, 1163, 177−189. (48) Gritti, G.; Guiochon, G. J. Chromatogr. A 2013, 1282, 58−71. (49) Environmental Protection Agency (EPA). 2010−2015 PFOA Stewardship Program. http://www.epa.gov/oppt/pfoa/pubs/ stewardship/. (50) Helfferich, F.; Peterson, D. L. Science 1963, 142, 661−663. (51) Ettre, L. S. Pure Appl. Chem. 1993, 65, 819−872. (52) Gilpin, R. K.; Jaroniec, M.; Lin, S. Anal. Chem. 1991, 63, 2849− 2852.

(53) Xu, Z.; Gibson, G. T. T.; Oleschuk, R. D. Analyst 2013, 138, 611− 619. (54) Xu, Z.; Oleschuk, R. D. J. Chromatogr. A 2014, 1329, 61−70. (55) Chester, T. L.; Coym, J. W. J. Chromatogr. A 2003, 1003, 101− 111. (56) Martin, A. J. P. Biochem. Soc. Symp. 1949, 3, 4−13. (57) Carr, P. W.; Li, J.; Dallas, A. J.; Eikens, D. I.; Tan, L. C. J. Chromatogr. A 1993, 656, 113−133. (58) Vailaya, A.; Horváth, C. J. Phys. Chem. 1997, 101, 5875−5888. (59) Weng, M.; Mallette, J.; Parcher, J. F. J. Chromatogr. A 2008, 1190, 1−7. (60) Samuelsson, J.; Arnell, R.; Diesen, J. S.; Tibbelin, J.; Paptchikhine, A.; Fornstedt, T.; Sjöberg, P. J. R. Anal. Chem. 2008, 80, 2105−2112. (61) Peterson, D. L.; Helfferich, F. J. Phys. Chem. 1965, 69, 1283−1293. (62) Sing, K. S. W. Pure Appl. Chem. 1982, 54, 2201−2218. (63) Kazakevich, Y. V.; LoBrutto, R.; Chan, F.; Patel, T. J. Chromatogr. A 2001, 913, 75−87. (64) Schay, G. Surface and Colloid Science; Wiley-Interscience: New York; Vol. 2, 1969; pp 155−212. (65) Undin, T.; Samuelsson, J.; Törncrona, A.; Fornstedt, T. J. Sep. Sci. 2013, 36, 1753−1761. (66) Vajda, P.; Cavazzini, A.; Felinger, A. J. Chromatogr. A 2010, 1217, 5965−5970. (67) Ratatunga, R. P. J.; Carr, P. W. Anal. Chem. 2000, 72, 5679−5692. (68) Handbook of Fluorous Chemistry; Gladysz, J. A., Curran, D. P., Horváth, I. T., Eds.; Wiley-VCH: New York, 2004.

4926

dx.doi.org/10.1021/ac500248b | Anal. Chem. 2014, 86, 4919−4926